Automated System for Detection of Cerebral Aneurysms in Medical CTA Images

摘要

In present scenario accurate detection of cerebral aneurysms in medical images plays a crucial role in reducing the incidents of subarachnoid hemorrhage (SAH) which carries a high rate of mortality. Many of the nontraumatic SAH cases are caused by ruptured cerebral aneurysms and accurate detection of these aneurysms can decrease a significant proportion of misdiagnosed cases. A scheme for automated detection of cerebral aneurysms is proposed in this study. The aneurysms are found by applying Normalization and generating the Probability Density Function (PDF) for the input image, local thresholding is used to identify appropriate aneurysm candidate regions. Feature vectors are calculated for the candidate regions based on gray-level, morphological and location based features. Rule based system is used to classify and detect cerebral aneurysms from candidate regions. Accuracy of the system is calculated using the sensitivity parameter.